







































All articles are permanently available online to the public without restrictions or subscription 

fees. All articles are free to be used, cited, and distributed, on condition that appropriate 

acknowledgment is included. Authors are the copyright holders of their original 

contributions and grant the Canadian Online Publication Group (COPG) a license to 

publish the article and identify itself as the original publisher. CPOJ articles are licensed 

under the Creative Commons Attribution 4.0 International License.   

 

CPOJ Website: https://jps.library.utoronto.ca/index.php/cpoj/index 

Editorial Office: cpoj@online-publication.com    

ISSN: 2561-987X 

CPOJ is a member of, and subscribes to the principles of, the Committee on Publication 

Ethics (COPE).   

 

 

VOLUME 5, ISSUE 1 

 2022 

 

RESEARCH ARTICLE 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 

Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

https://online-publication.com/wp/
http://creativecommons.org/licenses/by/4.0/
https://jps.library.utoronto.ca/index.php/cpoj/index
mailto:cpoj@online-publication.com
https://publicationethics.org/about/our-organisation
https://publicationethics.org/about/our-organisation
https://publicationethics.org/members/canadian-prosthetics-orthotics-journal
https://doi.org/10.33137/cpoj.v5i1.36744


 

1 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

 

 

 
RESEARCH ARTICLE 

 

EXPLORATION OF VIBROTACTILE BIOFEEDBACK STRATEGIES TO INDUCE STANCE TIME 

ASYMMETRIES 

Escamilla-Nunez R1,2 *, Sivasambu H2, Andrysek J1,2 

 

1 Institute of Biomedical Engineering, University of Toronto, Toronto, Canada. 
2 Bloorview Research Institute, Holland Bloorview Kids Rehabilitation Hospital, Toronto, Canada.  

 

 

 

 

  

 

 

 

 

 

 

 

 
 

 

 

 

INTRODUCTION   

Human gait is a complex physical activity involving the 

primary motor and somatosensory cortices, as well as  

the spinal cord (i.e., central pattern generator for 

locomotion), and the musculoskeletal system.1-3 The 

interaction  between  the  central  and  peripheral   nervous  

 

 

 

 

systems, reflexes, muscles, and joints allows individuals to 

ambulate in a stable, synchronized, and symmetrical 

manner.4 Gait symmetry is the degree of equality of 

biomechanical parameters between limbs within a gait 

cycle.4 Able-bodied gait is typically characterized by a high 

degree of symmetry. However, neurological disorders or 

physical impairments such as Parkinson’s,5 cerebral palsy,6 

stroke,7 incomplete spinal cord injury,8 and lower limb 

amputation,9 can lead to pathological gait, resulting in 

atypical and asymmetrical gait patterns.2 Gait asymmetry 

can affect diverse biomechanical and physiological 

 
OPEN  ACCESS Volume 5, Issue 1, Article No.2. 2022 

 

 

Journal Homepage: https://jps.library.utoronto.ca/index.php/cpoj/index 

 

ABSTRACT 

BACKGROUND: Gait symmetry is the degree of equality of biomechanical parameters between limbs 

within a gait cycle. Human gait is highly symmetrical; however, in the presence of pathology, gait often 

lacks symmetry. Biofeedback (BFB) systems have demonstrated the potential to reduce gait 

asymmetry, improve gait function, and benefit overall long-term musculoskeletal health. 

OBJECTIVE(S): The aim of this study was to develop a BFB system and evaluate three unique BFB 

strategies, including bidirectional control – constant vibration (BC), bidirectional control – variable 

vibration (BV), and unidirectional control – variable vibration (UV) relevant to gait symmetry. The 

assessed feedback strategies were a combination of vibration frequency/amplitude levels, vibration 

thresholds, and vibrotactile stimuli from one and two vibrating motors (tactors). Learning effect and 

short-term retention were also assessed. 

METHODOLOGY: Testing was performed using a custom BFB system that induces stance time 

asymmetries to modulate temporal gait symmetry. The BFB system continuously monitors specific gait 

events (heel-strike and toe-off) and calculates the symmetry ratio, based on the stance time of both 

limbs to provide real-time biomechanical information via the vibrating motors. Overall walking 

performance of ten (n=10) able-bodied individuals (age 24.8 ± 4.4 years) was assessed via metrics of 

symmetry ratio, symmetry ratio error, walking speed, and motor's vibration percentages.  

FINDINGS: All participants utilized BFB somatosensory information to modulate their symmetry ratio. 

UV feedback produced a greater change in symmetry ratio, and it came closer to the targeted symmetry 

ratio. Learning or short-term retention effects were minimal. Walking speeds were reduced with 

feedback compared to no feedback; however, UV walking speeds were significantly faster compared to 

BV and BC. 

CONCLUSION: The outcomes of this study provide new insights into the development and 

implementation of feedback strategies for gait retraining BFB systems that may ultimately benefit 

individuals with pathological gait. Future work should assess longer-term use and long-term learning 

and retention effects of BFB systems in the populations of interest.  

ARTICLE INFO 

Received: June 7, 2021 

Accepted: October 1, 2021 

Published: October 29, 2021 

CITATION 

Escamilla-Nunez R, Sivasambu H, 

Andrysek J. Exploration of 

vibrotactile biofeedback strategies 

to induce stance time asymmetries. 

Canadian Prosthetics & Orthotics 

Journal. 2022; Volume 5, Issue 1, 

No.2.https://doi.org/10.33137/cpoj.

v5i1.36744 

KEYWORDS 

Gait, Human Movement, 

Biofeedback, Learning Effect, 

Motor Control, Rehabilitation, 

Short-term Retention, Symmetry 

Ratio, Vibrotactile Feedback, 

Wearable Systems 

 

 

* CORRESPONDING AUTHOR 

Rafael Escamilla-Nunez  
Institute of Biomedical Engineering, University of Toronto, Toronto, 
Canada. 

Email: rafael.escamilla@mail.utoronto.ca 

ORCID ID: https://orcid.org/0000-0002-2739-878X 

https://doi.org/10.33137/cpoj.v5i1.36744
https://jps.library.utoronto.ca/index.php/cpoj/index
https://doi.org/10.33137/cpoj.v5i1.36744
https://doi.org/10.33137/cpoj.v5i1.36744
file:///D:/CPOJ/CPOJ%20Submitted%20Articles/ACCEPTED-Completed/86-Eighty%20six-Article-CPOJ-86-42A-2021-Jan-UoT-Canada/Article%20Proof/rafael.escamilla@mail.utoronto.ca
https://orcid.org/0000-0002-2739-878X


 

2 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
parameters.9 For instance, individuals with lower limb 

amputation often have reduced stance time support on the 

affected limb compared to the intact limb,9 as well as 

reduced walking speed, cadence,10 poor balance, and 

increased energy expenditure.11 Accordingly, the 

restoration of gait symmetry is critical for improving mobility, 

balance, function and efficiency, and overall long-term 

musculoskeletal health. Therefore, achieving gait symmetry 

is an important goal of gait rehabilitation. 

Gait rehabilitation typically entails motor learning and 

providing verbal cues related to the patient’s gait deviations 

or abnormal movement patterns to encourage positive 

changes. Gait rehabilitation is commonly provided by a 

physiotherapist, and the feedback is usually limited to 

subjective assessment of movement patterns. Additionally, 

rehabilitation sessions are often limited in duration and 

frequency.12 Patient barriers (e.g., long travel times, 

accessibility, etc.) and limited resources of healthcare 

facilities also often restrict access to physiotherapy.12  

Technology-driven approaches, such as therapy-focused 

videogames,13 virtual reality,14 and biofeedback (BFB),15,16 

have the potential to address the aforementioned 

challenges and provide alternative and augmentative 

means of training and rehabilitation in clinical settings or 

home-based environments. Specifically, BFB is the process 

of measuring physiological/biomechanical parameters and 

providing the user with real-time information about their 

current physical status.17 Wearable BFB systems for gait 

training can improve gait patterns by providing real-time, 

continuous feedback that reinforces good walking habits 

and physiotherapy goals.16,18  

One of the challenges of using BFB for rehabilitation is the 

establishment of effective feedback strategies and 

modalities (i.e., how biomechanical information is 

communicated to the BFB user).19,20 Compact and 

wearable auditory and visual BFB systems are available;16 

however, haptic BFB systems may be more suitable for field 

and community-based applications, since stimuli perception 

is less prone to be affected by external conditions such as 

noise or visual distractions.15 However, effectiveness of 

haptic modalities is highly dependent on the user’s ability to 

sense, interpret, and appropriately respond to the 

vibrotactile signals. Previous studies have demonstrated 

the effectiveness of vibrotactile-based BFB systems to alter 

gait and assessed the impact of varying properties such as 

vibration amplitude and frequency, location of tactors, 

interfaces, and pressures on somatosensory response.21-24 

However, few studies have applied a systematic approach 

to explore which BFB strategies most effectively achieve the 

desired symmetry targets. For instance, Afzal et al. tested 

different feedback strategies based on vibration durations 

and intensities, finding that greater alterations in gait 

symmetry occur with proportional vibrotactile feedback.25 

Lee et al. demonstrated that continuous vibration  

(i.e., progressive modulation of tactor's intensity) performed 

better than an ON/OFF vibration approach during dynamic 

weight-shifting balance training of elderly and individuals 

with Parkinson’s disease.19  

While substantial research has been conducted toward 

developing BFB strategies, to the best of the author’s 

knowledge, no study has attempted to compare and 

evaluate multiple vibrotactile biofeedback strategies based 

on gait symmetry targets, speed, and short-term learning 

effects when modulating gait symmetry. Hence, the 

overarching goal of this study was to develop a wearable 

vibrotactile BFB system and evaluate the effect of three 

unique BFB strategies on temporal gait symmetry and 

speed. In addition, aspects of learning and short-term 

retention effects were assessed by evaluating pre- and 

post-feedback gait parameters.  

METHODOLOGY 

A. System Instrumentation 

A BFB prototype system was developed that comprised of 

the following units (Figure 1). The Vibrating Unit (Figure 1a) 

included two vibrating motors (tactors) 9mm in diameter and 

25mm in length (model 307-103, Precision Microdrive, 

United Kingdom). Each vibrating motor was supplied with 

3.3V, corresponding to a nominal vibration frequency of 

250Hz and vibration amplitude of 7.5g (i.e., g = 9.8m/s2, the 

gravity of Earth). Recent studies suggest that higher 

frequencies (>230Hz), targeting Ruffini cylinders and 

Pacinian corpuscles skin mechanoreceptors, increase user 

detection accuracy and reduce reaction times after 

vibrotactile stimulation.22,23 The vibrating motors were 

adhered to the lower abdomen (using surgical tape, 

TransporeTM, 3M Canada) at the prolongation axis of the 

rectus femoris muscle following previous studies.26,27 The 

Microcontroller-based Control Unit (Figure 1b) was 

comprised of the Arduino Uno Rev3 (Sparkfun Electronics; 

Boulder, Colorado, USA) and a custom electronic board 

with N-type MOSFETs, diodes, and resistors, which was 

designed to ensure the correct operation and power supply 

of the vibrating motors and the Sensors/Transducer Unit. 

The Sensors/Transducer Unit (Figure 1c) included four-

square force sensitive resistors (FSRs) (model 406, Interlink 

Electronics, USA) to detect foot contact (heel-strike and toe-

off). The FSR’s force sensitivity ranged from 0.2N to 20N. 

Eight FSR sensors (four per foot) were adhered to the shoe 

sole, underneath the heel (x2) and the 1st and 5th 

metatarsal heads for the toe (x2). The Power Supply 

consists of a 5V at 5Ah Lithium-ion battery (PowerCore 

5000 by Anker Innovations, Shenzhen, China) powered the 

entire system. The Communication module included a 

Bluetooth serial communication device (HC-05 Bluetooth 

module by Smart Prototyping, Hong Kong) which provided 

wireless communication between the microcontroller and a 

host PC.  

https://doi.org/10.33137/cpoj.v5i1.36744


 

3 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 

An open-source software, Tera Term (Tera Term Project, 

Japan) was utilized for real-time data acquisition and 

visualization on the host PC at a sampling rate of 100 Hz 

(10ms resolution). A set-up of the BFB system on a 

participant is shown in Figure 1d. 

B. Biofeedback System Operation 

The BFB system employs a closed-loop design to 

continuously monitor specific gait events, namely heel-

strike (HS) and toe-off (TO) (Figure 2). Thus, the symmetry 

ratio (SR) was calculated to provide real-time 

biomechanical information via the vibrating motors to alter 

SR of BFB users. FSR thresholds for HS and TO onsets 

were determined by using a peak detection algorithm 

presented by Lopez-Meyer et al.28 The detection of HS and 

TO was used to compute the stance time (ST) of each leg. 

ST was defined as the amount of time that each leg remains 

in contact with the ground during each gait cycle.29,30 

Subsequently, SR was used to quantify gait symmetry 

based on the Equation:29,30 

𝑆𝑅 =  
𝑆𝑇𝑛𝑜𝑛−𝑑𝑜𝑚𝑖𝑛𝑎𝑛𝑡 𝑙𝑖𝑚𝑏

𝑆𝑇𝑑𝑜𝑚𝑖𝑛𝑎𝑛𝑡 𝑙𝑖𝑚𝑏
 

where, ST denotes the stance time of the non-dominant and 

dominant limbs, respectively.  SR was measured for each 

limb and used to provide feedback during the stance phase 

of the subsequent step.29,30 For non-pathological gait, SR 

values typically range between 0.95 to 1.05.25 Vibrotactile 

feedback was delivered to users based on the selected 

feedback strategies (i.e., a combination of vibration levels, 

vibration thresholds, and control algorithms for one and two 

motors activation strategies) as detailed below.  

C. Biofeedback Strategies 

Three novel strategies were applied that consisted of 

different vibration/amplitude levels, and number of stimuli 

(i.e. one or two motors). The three feedback strategies 

included 1) Bidirectional control – constant vibration (BC), 

2) Bidirectional control – variable vibration (BV), and 3) 

Unidirectional control – variable vibration (UV). 

Bidirectional control (BC and BV) provides feedback when 

the targeted SR value is either exceeded or not achieved 

(Figure 3a&b). Unidirectional control (UV) feedback only 

provides feedback if the targeted SR value is not achieved 

(Figure 3c). The unidirectional control uses a single vibrating 

motor, and bidirectional uses two motors (Figure 3). 

Vibrating motors are activated at two different vibration 

levels. A greater deviation from the targeted SR produces a 

vibration at 100% power, and a smaller error produces 

vibrations at 50% power. It should be noted that the 

magnitude of the SR error (i.e., the difference between the 

targeted SR and the currently measured SR), the pre-set 

vibration thresholds, and the targeted SR determine the 

 

Figure 1: Main components of the wearable BFB prototype (vibrating motors, microcontroller, Bluetooth, power supply, and FSRs sensors). 
(a) Vibrating Unit (motors) located at the lower abdomen at the prolongation axis of the rectus femoris muscle. (b) Microcontroller-based 
Control Unit, including the custom electronic board, the communication (Bluetooth) module, and the power supply. (c) Sensors/Transducer 
Unit comprises four FSRs sensors located at the heel (x2) and toe (x2) of each shoe sole. (d) Set-up of the BFB system on a participant. 

(a) (c) (d)

(b)

Microcontroller

Custom Electronic 

Board

Power Bank 

(5V, 5Ah)

Bluetooth 

Communication 

Module

FSR 

Sensors

Vibrating 

Motors   

https://doi.org/10.33137/cpoj.v5i1.36744


 

4 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
activation and vibration level of the motors according to the 

applied feedback strategy (Figure 3). A tolerance equal to 

0.05 was selected based on typical SR values for individuals 

with non-pathological gait (i.e., 0.95 ≤ SR ≤ 1.05, where 

SR=1.0 denotes perfect gait symmetry)25 (Figure 3). The UV 

strategy is designed to encourage the BFB user to move 

toward and exceed the SR target, at which point the 

vibrotactile feedback stops (Figure 3c). Strategies BC and 

BV require the BFB user to maintain SR within specified 

thresholds (Figure 3a&b).  

 

Figure 2: Control diagram of the BFB system to modulate symmetry ratio (SR). The closed-loop system includes system components, 

feedback strategies, and system operation. 

Biofeedback (BFB) User

Gait Cycle 

(measured for each limb)

Heel-strike 

(HS)

Toe-off 

(TO)
Heel-strike 

(HS)Stance Phase Swing Phase

Sensors/Transducer Unit

FSRs sensors

(Force Sensing Resistors) 

Custom Electronic  

Board

Four FSRs sensors per foot (heel x2 and toe x2) for heel-strike (HS) and toe-

off (TO) gait event detection

FSR for 

TO detection

FSR for 

HS detection

Feedback 

Strategies (FB)

Control 

Algorithm

Subject’s-specific control algorithm for HS and TO detection, including FSRs 

threshold and targeted SR. 

UV : Unidirectional control – variable vibration 

BV : Bidirectional control – variable vibration

BC : Bidirectional control – constant vibration

Microcontroller-based Control Unit

Vibrating Motors • One motor (UV)

• Two motors (BV, BC)

Vibration 

Level

50% and 100% of full frequency/amplitude 

level (Amplitude: 7.5g, Frequency: 250Hz)

Vibrating Unit

Vibration frequency/amplitude levels, vibration thresholds, including 

one and two motors vibration strategies.

HS and TO 

detection

Motor activation 

based on feedback 

and targeted SR

V
ib

ro
ta

c
ti

le
fe

e
d

b
a

c
k

 t
o

 a
lt

e
r 

g
a

it
 s

y
m

m
e

tr
y
 (

i.
e

.,
 s

y
m

m
e

tr
y
 r

a
ti

o
 –

S
R

) 
o

f 
B

F
B

 u
s

e
rs

ST and SR 

calculation

Microcontroller

Vibrating
motors

FSRs Sensors

Microcontroller

Vibrating
motors

FSRs Sensors

https://doi.org/10.33137/cpoj.v5i1.36744


 

5 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
 

 

B
id

ir
e
c
ti

o
n

a
l 
c
o

n
tr

o
l 

–
c
o

n
s
ta

n
t 

v
ib

ra
ti

o
n

 “
B

C
”

B
id

ir
e
c
ti

o
n

a
l 
c
o

n
tr

o
l 

–
v
a
ri

a
b

le
 v

ib
ra

ti
o

n
 “

B
V

”
U

n
id

ir
e
c
ti

o
n

a
l 
c
o

n
tr

o
l 
–

c
o

n
s
ta

n
t 

v
ib

ra
ti

o
n

 “
U

V
”

F
e
e
d

b
a
c
k
 

02

t

F
e
e
d

b
a
c
k

N
o

 

F
e
e
d

b
a
c
k

B
C

 F
e
e
d

b
a
c
k

T
a
rg

e
te

d
 S

RS
R

M
2

 @
 1

0
0

%

M
1

 @
 1

0
0

%

U
p

p
e

rT
H

R
*

L
o
w

e
rT

H
R

M
2

 @
 5

0
%

U
p

p
e

rT
H

R

M
1

 @
 5

0
%

M
1

 @
 1

0
0

%

M
2

 @
 1

0
0

%
F

e
e

d
b

a
c

k
 

02

t

F
e

e
d

b
a

c
k

N
o

 

F
e

e
d

b
a

c
k

B
V

 F
e
e
d

b
a
c
k

T
a
rg

e
te

d
 S

R

L
o
w

e
rT

H
R

 -
to

le
ra

n
c
e

S
R

L
o

w
e

rT
H

R

U
p

p
e

rT
H

R
 +

 t
o

le
ra

n
c
e

U
V

 F
e
e
d

b
a
c
k

M
1

 @
 5

0
%

M
1

 @
 1

0
0

%

02

t

F
e

e
d

b
a

c
k

N
o

 

F
e

e
d

b
a

c
k

T
a

rg
e

te
d

 S
R

L
o

w
e

rT
H

R

L
o
w

e
rT

H
R

-
T

o
le

ra
n
c
e
 

S
R

*
TH

R
 =

 T
h

re
sh

o
ld

N
o
 v

ib
ra

ti
o
n
 /

 N
o
 f

e
e
d
b
a
c
k
  

: 
L
o
w

e
rT

H
R

 ≤
 S

R
 ≤

  
U

p
p
e
rT

H
R

M
1
 @

 1
0
0
%

  
  

  
  

  
  

  
  

  
  

  
  

: 
S

R
 <

 L
o
w

e
r 

T
H

R

M
2
 @

 1
0
0
%

  
  

  
  

  
  

  
  

  
  

  
  

: 
S

R
 >

 U
p

p
e
r 

T
H

R

B
C

 =
 

N
o
 v

ib
ra

ti
o

n
 /
 N

o
 f

e
e

d
b

a
c
k
  

: 
L

o
w

e
rT

H
R

 ≤
 S

R
 ≤

  
U

p
p

e
rT

H
R

M
1

 @
 5

0
%

  
  
  
  
  
  
  

  
  
  
  
  
  
  
: 
L

o
w

e
rT

H
R

 –
to

le
ra

n
c
e

 <
 S

R
 <

 L
o

w
e

rT
H

R

M
2

 @
 5

0
%

  
  
  
  
  
  
  

  
  
  
  
  
  
  
: 
U

p
p

e
rT

H
R

 <
 S

R
 <

 U
p

p
e

rT
H

R
 +

 t
o

le
ra

n
c
e

M
1

 @
 1

0
0

%
  

  
  
  
  
  

  
  
  
  
  
  
  
: 
S

R
 <

 L
o

w
e

r 
T

H
R

 -
to

le
ra

n
c
e

M
2

 @
 1

0
0

%
  

  
  
  
  
  

  
  
  
  
  
  
  
: 
S

R
 >

 U
p
p

e
r 

T
H

R
 +

 t
o

le
ra

n
c
e

B
V

 =
 

N
o
 v

ib
ra

ti
o
n
 /

 N
o
 f

e
e
d
b
a
c
k
 :

 L
o
w

e
rT

H
R

 ≤
 S

R

M
1
 @

 1
0
0
%

  
  

  
  

  
  

  
  

  
  

  
 :
 L

o
w

e
rT

H
R

 –
to

le
ra

n
c
e
 <

 S
R

 <
 L

o
w

e
r 

T
H

R
  
  

  
  

  
  

  
  

  
  

  
  

  
  

  
  

 

M
1
 @

 5
0
%

  
  

  
  

  
  

  
  

  
  

  
  

 :
 S

R
 <

 L
o
w

e
rT

H
R

 –
to

le
ra

n
c
e

U
V

 =
 

(a
)

(b
)

(c
)

 

D
a
ta

 C
o

ll
e
c
ti

o
n

 S
e
s
s
io

n

S
ta

rt

T
ri

a
ls

B
F

B
 

S
e
tu

p

In
it

ia
l 
B

a
s
e

li
n

e

S
e
t 
F

S
R

 t
h
re

s
h
o
ld

, 
v
ib

ra
ti
o
n
 

th
re

s
h
o
ld

s
, 
a
n
d
 t
a
rg

e
te

d
 S

R

5
-1

0
m

in
 

x
 2

x
 2

x
 6

E
n

d

T
ri

a
ls

x
 2

N
o

 F
e

e
d

b
a

c
k

(N
F

)
N

o
 F

e
e
d

b
a
c
k

(N
F

)

R
a
n

d
o

m
iz

e
d

 F
e
e
d

b
a
c
k

(B
C

, 
B

V
, 
U

V
)

T
ra

in
in

g

(U
V

, 
B

C
, 
B

V
)

1
0
 m

in

x
 3

p
e
r 

c
o
n
d
it
io

n
b
re

a
k

F
ig

u
re

 3
: 

V
ib

ro
ta

c
ti
le

 f
e
e
d
b
a
c
k
 s

tr
a
te

g
ie

s
. 

(a
) 

B
C

: 
B

id
ir
e
c
ti
o

n
a
l 
c
o
n
tr

o
l 

–
 c

o
n
s
ta

n
t 

v
ib

ra
ti
o

n
; 

(b
) 

B
V

: 
B

id
ir
e
c
ti
o

n
a
l 

c
o
n
tr

o
l 

–
 v

a
ri
a

b
le

 v
ib

ra
ti
o

n
; 

a
n
d
 (

c
) 

U
V

: 
U

n
id

ir
e
c
ti
o

n
a
l 
c
o
n
tr

o
l 

–
 v

a
ri
a

b
le

 

v
ib

ra
ti
o

n
. 
B

C
 u

ti
liz

e
s
 t

w
o
 m

o
to

rs
 (

M
1

 &
 M

2
),

 w
h
ic

h
 p

ro
d
u
c
e
 v

ib
ra

ti
o
n
s
 a

t 
1
0
0
%

 o
f 

fu
ll 

v
ib

ra
ti
o

n
 f

re
q
u
e
n
c
y
/i
n

te
n
s
it
y
. 

B
V

 u
ti
li
z
e
s
 t

w
o
 m

o
to

rs
, 

w
h
ic

h
 p

ro
d
u
c
e
 v

ib
ra

ti
o

n
s
 a

t 
5
0
%

 a
n
d
 1

0
0
%

 o
f 

fu
ll 

v
ib

ra
ti
o

n
 f
re

q
u
e
n
c
y
/i
n
te

n
s
it
y
. 
A

n
d
, 
U

V
 u

ti
liz

e
s
 o

n
ly

 o
n
e
 m

o
to

r 
(M

1
),

 w
h
ic

h
 p

ro
d
u
c
e
s
 v

ib
ra

ti
o

n
s
 a

t 
5
0
%

 a
n
d
 1

0
0
%

 o
f 
fu

ll 
v
ib

ra
ti
o

n
 f
re

q
u
e

n
c
y
/i
n

te
n
s
it
y
. 
M

o
to

rs
 “

M
1

” 
a
n
d
 “

M
2

” 
w

e
re

 p
la

c
e
d
 a

t 
th

e
 

le
ft

 a
n
d
 r

ig
h
t 

s
id

e
 o

f 
th

e
 l

o
w

e
r 

a
b
d
o
m

e
n
 a

t 
th

e
 p

ro
lo

n
g
a
ti
o

n
 a

x
is

 o
f 

th
e
 r

e
c
tu

s
 f

e
m

o
ri
s
 m

u
s
c
le

, 
re

s
p
e
c
ti
v
e
ly

. 
V

ib
ra

ti
n

g
 t

h
re

s
h
o
ld

s
 a

re
 d

e
n
o
te

d
 b

y
 L

o
w

e
rT

H
R

 a
n
d
 U

p
p
e
rT

H
R

 f
o
r 

lo
w

e
r 

a
n
d
 

u
p
p
e
r 

th
re

s
h
o
ld

s
, 
re

s
p
e
c
ti
v
e
ly

. 
T

h
e
 d

e
s
ir
e
d
 S

R
 v

a
lu

e
 i
s
 d

e
n
o
te

d
 b

y
 T

a
rg

e
te

d
 S

R
. 
A

 t
o
le

ra
n
c
e
 o

f 
0
.0

5
 w

a
s
 s

e
le

c
te

d
 b

a
s
e
d
 o

n
 t
y
p

ic
a
l 
S

R
 v

a
lu

e
s
 f

o
r 

in
d
iv

id
u
a
ls

 w
it
h
 n

o
n
-p

a
th

o
lo

g
ic

a
l 
g
a
it
 (

i.
e
.,
 

0
.9

5
 ≤

 S
R

 ≤
 1

.0
5
, 
w

h
e
re

 S
R

=
1
.0

 d
e
n
o
te

s
 p

e
rf

e
c
t 
g
a
it
 s

y
m

m
e
tr

y
).

 

F
ig

u
re

 4
: 

 E
x
p
e
ri
m

e
n
ta

l 
p
ro

to
c
o
l 
fo

r 
th

e
 d

a
ta

 c
o
lle

c
ti
o

n
 s

e
s
s
io

n
. 
B

a
s
e
lin

e
 t
ri
a

ls
 c

o
n
s
is

t 
o
f 
w

e
a
ri
n

g
 t
h
e
 B

F
B

 s
y
s
te

m
, 
b
u
t 
n
o
 f
e
e
d
b
a
c
k
 (

N
F

) 
is

 p
ro

v
id

e
d
. 
F

e
e
d
b
a
c
k
 s

tr
a
te

g
ie

s
 (

B
C

, 
B

V
, 
U

V
) 

w
e
re

 

ra
n
d
o
m

iz
e
d
 f
o
r 

d
a
ta

 c
o
lle

c
ti
o

n
. 
B

C
: 

 B
id

ir
e
c
ti
o

n
a
l 
c
o
n
tr

o
l 
–
 c

o
n
s
ta

n
t 
v
ib

ra
ti
o

n
, 
B

V
: 
B

id
ir
e
c
ti
o

n
a
l 
c
o
n
tr

o
l 
–
 v

a
ri
a

b
le

 v
ib

ra
ti
o

n
, 

a
n
d
 U

V
: 

U
n
id

ir
e
c
ti
o

n
a
l 
c
o
n
tr

o
l 
–
 v

a
ri
a

b
le

 v
ib

ra
ti
o
n
. 

https://doi.org/10.33137/cpoj.v5i1.36744


 

6 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
Previous studies suggest that permanent gait changes must 

be achieved gradually,30 finding that absolute differences 

larger than 5% can be considered as potential improvement 

or deterioration of gait symmetry.31 For this reason, all 

feedback strategies (BC, BV, UV) targeted SRs that were 

10% greater than an initial SR baseline (e.g., mean initial 

SR baseline = 1.0; targeted SR = 1.10), as proposed by.25 

SR converging towards 1, indicates an improvement in gait 

symmetry, whereas SR diverging from 1, indicates a 

deterioration.31 Since this study involved participants with 

non-atypical gait symmetry (i.e., SR = 1 ± 0.05), the testing 

paradigm consisted of deviating SR from SR = 1 to alter gait 

symmetry. For all the feedback strategies, vibrations (if 

provided) start at HS and end at TO of the same limb within 

a gait cycle. Vibrations are only provided if participants are 

walking with a SR value outside of the pre-set vibrating 

thresholds (Figure 3).  

D. Participants 

The study involved a convenience sample of ten (n=10) 

healthy subjects (five males), age 24.8 ± 4.4yrs; height 1.7 

± 0.1m; weight 68.7 ± 14.4kg. Participants were 18 years or 

older, were all English speaking, and having no physical or 

gait-related impairments, ambulation difficulties or neuro-

motor disorders. The study was approved by the Research 

Ethics Board (REB #16-675) at Holland Bloorview Hospital, 

Canada. Informed written consent was obtained from each 

participant before commencing. 

E. Experimental Protocol 

Data were collected in a single session. Participants were 

instrumented with the BFB system (Figure 1). Training was 

provided before collecting data (Figure 4). Training 

consisted of a brief explanation about the BFB operation 

and the opportunity to walk using each feedback strategy 

(BC, BV, UV). During training, participants were coached 

about how to interpret the vibrotactile feedback. To 

determine limb dominance, participants were asked about 

what foot they use to kick a ball. All participants were right 

footed (i.e., right limb was the dominant limb). For the UV 

strategy, the vibrating motor (M1) was placed on the non-

dominant (left) side at the lower abdomen level. For the BV 

and BC strategies, motor M1 was placed on the non-

dominant (left) side and the second motor (M2) on the 

dominant (right) side at the lower abdomen. Since the goal 

was to achieve a 10% change in SR (i.e., increase ST on 

the non-dominant “left” limb), verbal instruction and cues 

were provided as follow: for UV strategy (motor M1 placed 

on the non-dominant side), “if the motor on your left side 

vibrates, you need to spend more time in contact with the 

ground on that (left) side”. Hence, for BV and BC strategies, 

a verbal instruction/cue for motor M2 was provided in 

addition to the one provided for M1, “if the motor on your 

right side vibrates, you are spending too much time on your 

left side, so you just need to spend a little bit less time in 

contact with the ground on your left side”. In terms of the 

vibration levels, cues consisted of “while walking, you will 

experience two different vibration intensities, the weaker 

vibration means you are closer to the target, and stronger 

means you are farther from the target. The goal is to receive 

no vibration”. For the data collection, each participant 

performed 30 walking trials in total (i.e., 4 No Feedback 

trials plus 6 Feedback trials for each of the three strategies). 

Each trial consisted of walking 20 meters in a straight line at 

a self-selected speed. Feedback strategies (BC, BV, UV) 

were randomized using simple and balanced randomization 

through a random number generator. No feedback “NF” 

condition was performed before (x2 trials) and after (x2 

trials) each feedback strategy. Both NF conditions (before 

and after feedback) were used as a baseline to compare 

BFB effects within strategies. Average gait speed was 

calculated for each trial based on the walking distance and 

time recorded with a stopwatch. The magnitude of the SR 

error was calculated for each gait cycle as the difference 

between the targeted SR and the measured SR. The 

percentage of vibration was calculated based on the 

activation status of each motor, which indicates the number 

of times that the motors were activated (ON = 1) or 

deactivated (OFF = 0) for each trial (i.e., time (ON/(ON + 

OFF))*100). 

F. Data Analysis 

Data recorded and captured using TeraTerm software (Tera 

Term Project, Japan) were exported to Excel (Microsoft 

Corp; Redmond, Washington) and processed in MATLAB 

2019b (R2019b, Mathworks, MA, USA) to extract parameter 

values (i.e., SR, SR error, average gait speed, and 

percentage of vibration) for each trial, condition, and 

participant. Statistical analysis was performed using JMP 

Pro 2019 software (Statistical Discovery, SAS, USA). A 

Shapiro-Wilk W Test with an alpha level of 0.05 was used 

to confirm the assumption of normal distribution of the data. 

A multivariate analysis of variance (MANOVA) was 

performed between and within participants, trials, and 

conditions (NF, BC, BV, UV) across all retrieved 

parameters. Statistical significance was determined using a 

critical alpha level of 0.05 for all primary analyses. If 

statistically significant differences were found, a Fit Model – 

Mixed Model analysis with a post-hoc Tukey HSD all 

pairwise comparisons analysis was performed to identify 

which particular differences between pairs of means were 

significant. In addition, a paired t-test was performed to 

compare differences on the level of precision with which the 

targeted SR was achieved between conditions (NF, BC, BV, 

UV). For this test, the standard deviation of the SR values 

was used. Finally, a paired t-test was used to compare SR 

values between NF conditions (i.e., NF before and after 

vibrotactile feedback) to identify statistically significant 

short-term retention effects. A Bonferroni correction with an 

https://doi.org/10.33137/cpoj.v5i1.36744


 

7 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
adjusted critical alpha level of 0.008 (p = 0.05/6) was 

applied to reduce type I errors for multiple pairwise 

comparisons. 

RESULTS 

A. Biofeedback Strategies Effectiveness 

Changes on SR values were found statistically significant 

when comparing between the feedback strategies (BC, BV, 

UV) and the no feedback (NF) condition (p < .001). 

Significant differences were also found within feedback 

strategies. UV produced a larger change in SR than BV (p 

< .001) and BC (p < .001) (Figure 5). However, changes in 

SR values between strategies BV and BC were not 

significantly different (p = 0.708). In terms of the SR error, 

all BFB strategies resulted in larger SR errors that were 

statistically different from NF (NF–BC: p < .001; NF–BV: p 

< .001; and NF–UV: p < .001) (Figure 6). There were also 

statistically significant differences among the SR errors for 

BC–UV (p < .001) and BV–UV (p < .001) strategies, but not 

among BC–BV (p = 0.512). Further, when comparing the 

standard deviation of SR, NF achieved a higher level of 

precision (smaller variability) compared than all the 

feedback strategies (p < .001; BC, BV, UV). However, no 

significant differences in the level of precision (i.e., standard 

deviation of SR) were found between feedback strategies 

(BC-UV: p = 0.057; BC–BV: p = 0.065; and BV–UV: p = 

0.752). 

B. Short-term Retention and Learning Effects 

In terms of short-term retention, no significant effects  

(p = 0.156) were evident from Figure 7, and based on the 

paired t-test statistical analysis of NF before (i.e., NF1) and 

after (i.e., NF2) vibrotactile feedback (BC, BV, UV). 

Learning effects for trial*feedback interactions within 

subjects were not statistically significant (Figure 8). 

However, for BV and UV strategies, participants neared the 

targeted SR from the first trial, with a slight but non-

significant trend towards improvement in subsequent trials 

(Figure 8).   

C. Secondary Outcomes 

All feedback conditions were associated with significantly 

slower walking speeds compared to the NF condition which 

was 1.398 m/s ± 0.022 (p < .001). When comparing walking 

speed between feedback strategies (BC, BV, UV), 

participants walked significantly faster with UV feedback 

(1.206 ± 0.029 m/s, p < .001) than BV (1.067 ± 0.023 m/s, 

p < .001) and BC (1.053 ± 0.020 m/s, p < .001). No 

significant differences in walking speeds were found 

between BV and BC (p = 0.900).  

 

Figure 5: SR vs FB. Box plot of SR values for all feedback (BC, BV, and UV) and no feedback (NF) conditions. BC:  Bidirectional control – 

constant vibration; BV: Bidirectional control – variable vibration; UV: Unidirectional control – variable vibration. Statistically significant 

differences between conditions are denoted with an ‘*’. Mean, standard deviations (Std Dev), maximum, median, and minimum values of SR 

are presented. Upper and Lower Thresholds labeled as Upper_Thresh and Lower_Thresh, respectively. 

NF BC BV UV

Feedback Condition

SR across Feedback Conditions

S
y
m

m
e

tr
y
 R

a
ti

o
 (

S
R

)

Baseline = 1.0

0.95

1.20

1.25

1.30

Target_SR = 1.10

Upper_Thresh = 1.15

Lower_Thresh = 1.05

Mean

Std Dev

Max

Median

Min

0.995

0.016

1.040

1.000

0.960

1.086

0.037

1.190

1.080

1.000

1.083

0.030

1.180

1.080

1.010

1.104

0.043

1.200

1.100

1.000

*
*

*

*
*

https://doi.org/10.33137/cpoj.v5i1.36744


 

8 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 6: SR Error. Mean SR error values across feedback (BC, BV, UV) and no feedback (NF) conditions. BC:  Bidirectional control – 

constant vibration; BV: Bidirectional control – variable vibration; UV: Unidirectional control – variable vibration. Statistically significant 

differences were found between NF and feedback conditions (BC, BV, and UV), including between feedback strategies UV-BC and UV-BV, 

but not between BC – BV strategies. Statistically significant differences are denoted with an ‘*’. Mean and standard deviations (Std Dev) are 

listed above. 

SR Error across Feedback Conditions

NF UV

Feedback Conditions

BC BV

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

0.12

0.14
S

R
 E

rr
o

r

-0.10

0.08

0.10

0.12

Mean

Std Dev

0.005

0.016
0.014

0.037

0.017

0.030

-0.004

0.043
*

*
*

*
*

 

Figure 7: Short-term Retention.  Box plot of SR values during No Feedback (NF) before (i.e., NF1) and after (i.e., NF2) providing vibrotactile 

feedback (BC, BV, UV). BC:  Bidirectional control – constant vibration; BV: Bidirectional control – variable vibration; UV: Unidirectional control 

– variable vibration. Statistically significant differences were found only between NF and feedback conditions. Mean and standard deviations 

(Std Dev) are also included above. Upper and Lower Thresholds labeled as Upper_Thresh and Lower_Thresh, respectively. 

0.95

NF1 FB NF2

UVBVBC

Short-term Retention (NF vs FB)

SR Target = 1.10

Lower_Thresh = 1.05

Upper_Thresh = 1.15

Baseline = 1.0

S
y
m

m
e
tr

y
 R

a
ti

o
 (

S
R

)

1.20

Mean

Std Dev

0.992

0.013

1.086

0.037

0.997

0.021

0.996

0.016

1.083

0.030

0.996

0.015

0.994

0.016

1.104

0.043

0.995

0.016

Feedback Conditions

NF1 NF1FB FBNF2 NF2

https://doi.org/10.33137/cpoj.v5i1.36744


 

9 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 

The percentage of vibration was significantly smaller under 

UV feedback (21%) compared to BV (27%), and BC (52%). 

The post-hoc Tukey HSD analysis showed a significant 

vibration percentage difference between conditions BC and 

UV (p < .001) and, BC and BV (p < .001). There was no 

significant difference in the vibration percentage between 

UV and BV (p = 0.959). 

DISCUSSION 

The purpose of this study was to investigate the effect of 

BFB on SR and walking speed via three novel vibrotactile 

feedback strategies (BC, BV, UV), which are different 

combinations of vibration thresholds, vibration levels 

(frequencies/amplitudes), and control strategies based on 

the activation of one and two motors. These feedback 

strategies were utilized to provide somatosensory 

information to BFB users to modulate gait symmetry during 

walking. In addition, learning effects and short-term 

retention were investigated. 

During walking trials, all participants were able to utilize BFB 

somatosensory information to alter their gait performance 

towards the targeted symmetry ratios (SR). Accordingly, 

results showed that BC, BV, and UV feedback strategies 

can all potentially modulate SR of BFB users.  

The results also suggest that a unidirectional strategy (UV) 

can produce a greater change in SR, to bring it closer to the 

target value (lower SE error). Hence, UV more accurately 

achieved the target SR as compared to both BC and BV. 

While the precision (variability) was not significantly 

different among feedback strategies, based on Figure 5 and 

Figure 6 the bidirectional strategies (BC and BV) trended 

toward more precise changes in SR (i.e., error bars are 

larger for UV compared to BC and BV). This may be due to 

the target being exceeded more frequently (i.e., SR higher 

than set target). In contrast, the bidirectional strategies (BC 

and BV) having both an upper and lower limit around the 

target values, produced less variability. It must be noted that 

among all of the conditions, participants achieved the lowest 

SR variability in the NF condition. One approach to achieve 

both accuracy and precision would be to use a bidirectional 

strategy with an adaptive targeted SR control, by which the 

target SR is gradually increased as the user changes their 

SR. This technique may reduce variability by guiding users 

to perform smaller step-to-step increments.   

 

Figure 8: Learning Effects. Box plot of SR values for mean SR values of all participants across trials during No Feedback (NF) and Feedback 

conditions (BC, BV, UV). BC:  Bidirectional control – constant vibration; BV: Bidirectional control – variable vibration; UV: Unidirectional control 

– variable vibration. No statistically significant differences were found among conditions. Mean and standard deviations (Std Dev) are also 

included above for each trial. Upper and Lower Thresholds labeled as Upper_Thresh and Lower_Thresh, respectively. 

Upper_Thresh = 1.15

Lower_Thresh = 1.05

Target SR = 1.10 

S
y
m

m
e

tr
y
 R

a
ti

o
 (

S
R

)

Baseline = 1.0

0.95

1.20

1.25

1.30

NF BC BV UV

Feedback Conditions

SR across Trials and Feedback Conditions                   

(Trial 1)

(Trial 2)

(Trial 3)

(Trial 4)

(Trial 5)

(Trial 6)

Mean

0.990

0.997

0.995

0.996

0.997

0.995

Mean

1.087

1.091

1.080

1.080

1.101

1.080

Mean

1.084

1.081

1.076

1.081

1.089

1.089

Mean

1.094

1.105

1.099

1.103

1.114

1.111

Std Dev

0.016

0.016

0.013

0.020

0.016

0.015

Std Dev

0.039

0.042

0.029

0.035

0.037

0.036

Std Dev

0.033

0.032

0.031

0.026

0.030

0.028

Std Dev

0.039

0.047

0.043

0.040

0.046

0.041

Trial 1 

Trial 2 

Trial 3 

Trial 4 

Trial 5 

Trial 6 

 

 

 

 

 

 

https://doi.org/10.33137/cpoj.v5i1.36744


 

10 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
Few differences were evident between the variable and 

constant bidirectional strategies (i.e., BV and BC, 

respectively), suggesting limited effectiveness in using 

vibration levels for thresholding. In this study the two distinct 

levels of vibration (both frequency and amplitude) were 

applied based on the magnitude of the error and pre-set 

vibration thresholds. Previous studies have resorted to 

altering duration of vibrations;27,30,32,33 and, only few of 

them have used variable amplitudes.19,25 In terms of the 

effectiveness of the vibration pattern (i.e., continuous 

versus corrective versus ON/OFF or discrete feedback), the 

previous research findings are mixed. Some studies 

suggest that continuous feedback (i.e., progressively 

incrementing or decrementing motor's intensity) produce 

greater gait improvements over discrete feedback (i.e., 

motors ON/OFF activation);19 and vice versa.27,34 Whereas, 

others suggest that corrective feedback (e.g., vibration only 

if targeted value is not reached) can elicit greater effects 

compared to continuous feedback (e.g., vibration until 

targeted value is reached).32 The present study combines 

discrete signals (ON/OFF) with corrective feedback, adding 

multiple vibration levels and thresholds, which provides the 

BFB system with unique feedback strategies to modulate 

gait symmetry of BFB users.  

In this study we found the BFB learning effects and short-

term retention to be minimal. According to the literature, 

learning a new skill or eliciting a locomotor adaptation is a 

complex process that involves motor adaptation, skill 

acquisition, and decision-making.35,36 The process of motor 

learning occurs gradually and improves over time. The 

learning process, at early stages, demands high cognitive 

effort, high consciousness of the task performed, and 

greater amounts of energy. However, at later stages, the 

movements seem to occur more unconsciously, 

automatically, and with less effort.36 It is plausible that 

extended use of the BFB system may result in relearning, 

and retention whereby the modified gait patterns would be 

preserved once the BFB system is no longer active. For 

instance, a recent study showed that 1 out of 3 above-knee 

amputee participants were able to retain improvements in 

gait symmetry (+14.9% improvement compared to baseline) 

after three training sessions of using vibrotactile feedback, 

suggesting an effective motor learning at least in the short-

term.37 

The slower than normal walking speeds associated with the 

provision of BFB, indicate potential limitations in terms of 

BFB effectiveness. However, it is foreseeable, that over the 

longer term as users utilize BFB less consciously, walking 

speeds may naturally recover towards normal values. 

Moreover, when feedback was provided, UV feedback 

achieved a significantly faster walking speed compared to 

BV and BC. UV also resulted in less vibrotactile feedback 

(vibration percentage) compared to BV and BC. Together, 

these results may indicate that as the complexity of the 

feedback strategy and information provided to BFB user 

increases, thus taxing of the executive function (i.e., 

cognitive processes), slower execution of function (e.g., 

slower motor response to stimulation) results. It might be 

beneficial for new BFB users to start the gait retraining with 

UV feedback. Once UV is learned, treatment can move 

progressively to BV or BC strategies. 

Future iterations of the BFB system should incorporate 

visual or auditory feedback modalities to assess the 

effectiveness of multimodal feedback paradigms. In 

addition, future studies should have in mind that individuals 

with poor somatosensory function might have decreased 

sensory perception to stimulus detection, which might affect 

the performance of haptic BFB systems. However, the 

stochastic resonance phenomenon, which consists of 

delivering sub-threshold noise to the somatosensory 

system might be a promising alternative for enhancing 

sensitivity to sensory inputs,38 and for improving reaction 

times.39 Accordingly, traditional haptic BFB systems can be 

combined with a noise-generating device/module to 

enhance BFB sensory perception to improve gait 

asymmetries. Also, the effects of haptic BFB systems on the 

tonic vibration reflexes (i.e., reflex muscular contraction) 

and the excitatory and inhibitory responses of the muscle 

spindle, which play a role enhancing muscle activation 

should be further investigated, since acute indirect 

vibrations acting on muscles can potentially enhance force, 

power, flexibility, balance, and proprioception, which might 

suggest neural enhancement.40  

This study has several limitations. The limited sample size 

of healthy subjects with non-asymmetrical gait represented 

a main limitation for generalizable conclusions. Thus, the 

performance of the BFB system in populations with 

pathological gait remains to be studied. Additionally, 

pressure sensors had a tendency to degrade over time 

resulting in inconsistent measurements. In such instances 

where measurements became unreliable, sensors were 

replaced, and additional data collected within the session. 

However, the development of a clinically relevant system 

will require more robust sensing instrumentation. To 

address the measurement issues with the FSRs, and also 

improve wearability, inertial measurement units should be 

considered. A final potential limitation relates to the sound 

generated by the vibrating motors. It is possible that it may 

have contributed to the feedback received by the 

participants. Thus, using headphones to cancel external 

sources of noise might be a point of consideration for future 

studies involving haptic feedback. As part of future work, 

secondary/indirect changes in gait patterns (either 

improvements or detriments) due to BFB should be 

investigated. Finally, longer-term use of BFB is needed to 

assess learning and retention effects. 

 

https://doi.org/10.33137/cpoj.v5i1.36744


 

11 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
CONCLUSION 

In this paper, the development of a wearable vibrotactile 

BFB system was presented along with the evaluation of 

three novel feedback strategies to modulate temporal gait 

symmetry by inducing stance time asymmetries. Clinical 

testing of the BFB prototype showed its ability to alter SR 

during walking; however, no learning effects or short-term 

retention effects were found.  

ACKNOWLEDGEMENTS 

The authors would like to thank Össur hf. for their continuous 

support and for reviewing the manuscript. 

DECLARATION OF CONFLICTING 

INTERESTS 

The authors declare no conflict of interest. The funders had no role 

in the design of the study; in the collection, analyses, or 

interpretation of data; in the writing of the manuscript, or in the 

decision to publish the results. 

AUTHOR CONTRIBUTION 

• Rafael Escamilla-Nunez: Conceptualization, methodology, 

validation, formal analysis, investigation, writing—original 

draft preparation, writing/review and editing, visualization, 

funding acquisition. 

• Harry Sivasambu: Validation, formal analysis, investigation, 

writing/review and editing, visualization. 

 

• Jan Andrysek: Supervision Conceptualization, methodology, 

investigation, writing/review and editing. 

SOURCES OF SUPPORT 

This research was funded by Natural Sciences and Engineering 

Research Council of Canada (NSERC) Discovery RGPIN 2018-

05046, NSERC CRD CRDPJ 491125 – 15, and Mexico’s National 

Council for Science and Technology (CONACyT). 

ETHICAL APPROVAL 

The study was approved by the Research Ethics Board (REB #16-

675) at Holland Bloorview Hospital, Canada. Informed written 

consent was obtained from each participant before commencing.  

REFERENCES 

1.Kaas JH, Merzenich MM, Killackey HP. The reorganization of 

somatosensory cortex following peripheral nerve damage in adult 

and developing mammals. Annu Rev Neurosci. 1983;6:325–56. 

DOI:10.1146/annurev.ne.06.030183.001545 

2. Eshraghi A., Andrysek J. (2018) Influence of Prosthetic Socket 

Design and Fitting on Gait. In: Müller B., Wolf S. (eds) Handbook of 

Human Motion. Springer, Cham. Pages 1383-1406. DOI: 10.1007/ 

978-3-319-14418-4_76 

3.Guertin PA. Central pattern generator for locomotion: anatomical, 

physiological, and pathophysiological considerations. Front Neurol. 

2013;3:1–15. DOI:10.3389/fneur.2012.00183 

4.Sadeghi H, Allard P, Prince F, Labelle H. Symmetry and limb 

dominance in able-bodied gait: a review. Gait Posture. 2000;12:34–

45. DOI:10.1016/S0966-6362(00)00070-9 

5.Shen X, Mak MKY. Balance and gait training with augmented 

feedback improves balance confidence in people with parkinson’s 

disease. Neurorehabil Neural Repair. 2014;28:524–35. 

DOI:10.1177/1545968313517752 

6.MacIntosh A, Lam E, Vigneron V, Vignais N, Biddiss E. 

Biofeedback interventions for individuals with cerebral palsy: a 

systematic review. Disabil Rehabil. 2019;41:2369–91. DOI: 

10.1080/09638288.2018.1468933 

7.Afzal MR, Pyo S, Oh M-K, Park YS, Lee B-C, Yoon J. Haptic 

based gait rehabilitation system for stroke patients. In 2016 

IEEE/RSJ Int Conf Intell Robot Syst. 2016;3198–203. 

DOI:10.1109/IROS.2016.7759494 

8.Thrasher TA, Flett HM, Popovic MR. Gait training regimen for 

incomplete spinal cord injury using functional electrical stimulation. 

Spinal Cord. 2006;44:357–61. DOI:10.1038/sj.sc.3101864 

9.Sagawa Y, Turcot K, Armand S, Thevenon A, Vuillerme N, 

Watelain E. Biomechanics and physiological parameters during gait 

in lower-limb amputees: a systematic review. Gait Posture. 

2011;33:511–26. DOI:10.1016/j.gaitpost.2011.02.003 

10.Nolan L, Wit A, Dudziñski K, Lees A, Lake M, Wychowañski M. 

Adjustments in gait symmetry with walking speed in trans-femoral 

and trans-tibial amputees. Gait Posture. 2003;17:142–51. 

DOI:10.1016/S0966-6362(02)00066-8 

11.Miller WC, Deathe AB, Speechley M, Koval J. The influence of 

falling, fear of falling, and balance confidence on prosthetic mobility 

and social activity among individuals with a lower extremity 

amputation. Arch Phys Med Rehabil. 2001;82:1238–44. DOI: 

10.1053/apmr.2001.25079 

12.Seymour R. Prosthetics and orthotics: lower limb and spinal. 

Lippincott Williams & Wilkins; 2002. 

13.Staiano AE, Flynn R. Therapeutic uses of active videogames: a 

systematic review. Games Health J. 2014;3:351–65. DOI:10.1089/ 

g4h.2013.0100 

14.Darter BJ, Wilken JM. Gait training with virtual reality–based 

real-time feedback: improving gait performance following 

transfemoral amputation. Phys Ther. 2011;91:1385–94. 

DOI:10.2522/ptj.20100360 

15.Van Gelder LMA, Barnes A, Wheat JS, Heller BW. The use of 

biofeedback for gait retraining: A mapping review. Clin Biomech. 

2018;59:159–66. DOI:10.1016/j.clinbiomech.2018.09.020 

16.Escamilla-Nunez R, Michelini A, Andrysek J. Biofeedback 

systems for gait rehabilitation of individuals with lower-limb 

amputation: a systematic review. Sensors. 2020;20:1628. 

DOI:10.3390/s20061628 

17.Giggins OM, Persson U, Caulfield B. Biofeedback in 

rehabilitation. J Neuroeng Rehabil. 2013;10:60. DOI:10.1186/1743-

0003-10-60 

https://doi.org/10.33137/cpoj.v5i1.36744


 

12 

Escamilla-Nunez R, Sivasambu H, Andrysek J. Exploration of vibrotactile biofeedback strategies to induce stance time asymmetries. Canadian Prosthetics & 
Orthotics Journal. 2022; Volume 5, Issue 1, No.2. https://doi.org/10.33137/cpoj.v5i1.36744 

ISSN: 2561-987X 
VIBROTACTILE BIOFEEDBACK  

Escamilla-Nunez et al., 2022 CPOJ 

 
18.Shull PB, Damian DD. Haptic wearables as sensory 

replacement, sensory augmentation and trainer – a review. J 

Neuroeng Rehabil. 2015;12:59. DOI:10.1186/s12984-015-0055-z 

19.Lee B-C, Fung A, Thrasher TA. The effects of coding schemes 

on vibrotactile biofeedback for dynamic balance training in 

parkinson’s disease and healthy elderly individuals. IEEE Trans 

Neural Syst Rehabil Eng. 2018;26:153–60. DOI:10.1109/ 

TNSRE.2017.2762239 

20.Chamorro-Moriana G, Moreno A, Sevillano J. Technology-

based feedback and its efficacy in improving gait parameters in 

patients with abnormal gait: a systematic review. Sensors. 

2018;18:142. DOI:10.3390/s18010142 

21.Shi S, Leineweber MJ, Andrysek J. Exploring the tactor 

configurations of vibrotactile feedback systems for use in lower-limb 

prostheses. J Vib Acoust. 2019;141. DOI:10.1115/1.4043610 

22.Sharma A, Leineweber MJ, Andrysek J. Effects of cognitive load 

and prosthetic liner on volitional response times to vibrotactile 

feedback. J Rehabil Res Dev. 2016;53:473–82. DOI:10.1682/ 

JRRD.2015.04.0060 

23.Sharma A, Torres-Moreno R, Zabjek K, Andrysek J. Toward an 

artificial sensory feedback system for prosthetic mobility 

rehabilitation: examination of sensorimotor responses. J Rehabil 

Res Dev. 2014;51:907–17. DOI:10.1682/JRRD.2013.07.0164 

24.Leineweber MJ, Shi S, Andrysek J. A Method for evaluating 

timeliness and accuracy of volitional motor responses to vibrotactile 

stimuli. J Vis Exp. 2016;2016. DOI:10.3791/54223 

25.Afzal MR, Lee H, Eizad A, Lee CH, Oh M-K, Yoon J. Evaluation 

of novel vibrotactile biofeedback coding schemes for gait symmetry 

training. 2019 2nd IEEE Int Conf Soft Robot. 2019, 540–5. 

DOI:10.1109/ROBOSOFT.2019.8722751 

26.Lee B-C, Thrasher TA, Fisher SP, Layne CS. The effects of 

different sensory augmentation on weight-shifting balance 

exercises in Parkinson’s disease and healthy elderly people: a 

proof-of-concept study. J Neuroeng Rehabil. 2015;12:75. 

DOI:10.1186/s12984-015-0064-y 

27.Crea S, Edin BB, Knaepen K, Meeusen R, Vitiello N. Time-

discrete vibrotactile feedback contributes to improved gait 

symmetry in patients with lower limb amputations: case series. 

Phys Ther. 2017;97:198–207. DOI:10.2522/ptj.20150441 

28.Lopez-Meyer P, Fulk GD, Sazonov ES. Automatic detection of 

temporal gait parameters in poststroke individuals. IEEE Trans Inf 

Technol Biomed. 2011; 15:594–601. DOI:10.1109/TITB.2011. 

2112773 

29.Patterson KK, Gage WH, Brooks D, Black SE, McIlroy WE. 

Evaluation of gait symmetry after stroke: a comparison of current 

methods and recommendations for standardization. Gait Posture. 

2010;31:241–6. DOI:10.1016/j.gaitpost.2009.10.014 

30.Redd CB, Bamberg SJM. A wireless sensory feedback device 

for real-time gait feedback and training. IEEE/ASME Trans 

Mechatronics. 2012;17:425–33. DOI:10.1109/TMECH.2012. 

2189014 

31.Pagel A, Arieta AH, Riener R, Vallery H. Effects of sensory 

augmentation on postural control and gait symmetry of 

transfemoral amputees: a case description. Med Biol Eng Comput. 

2016;54:1579–89. DOI:10.1007/s11517-015-1432-2 

32.Plauche A, Villarreal D, Gregg RD. A haptic feedback system for 

phase-based sensory restoration in above-knee prosthetic leg 

users. IEEE Trans Haptics. 2016;9:421–6. DOI:10.1109/ 

TOH.2016.2580507 

33.Maldonado-Contreras J, Marayong P, Khoo I-H, Rivera R, Ruhe 

B, Wu W. Proprioceptive improvements of lower-limb amputees 

under training with a vibrotactile device — a pilot study. 2017 IEEE 

Healthc Innov Point Care Technol. 2017;229–32. DOI: 

10.1109/HIC.2017.8227626 

34.Crea S, Cipriani C, Donati M, Carrozza MC, Vitiello N. Providing 

time-discrete gait information by wearable feedback apparatus for 

lower-limb amputees: usability and functional validation. IEEE 

Trans Neural Syst Rehabil Eng. 2015;23:250–7. DOI: 

10.1109/TNSRE.2014.2365548 

35.Kitago TO, Krakauer JW. Motor learning principles for 

neurorehabilitation. Handb Clin Neurol. 2013;110:93-103. DOI: 

10.1016/B978-0-444-52901-5.00008-3 

36.Cech DJ, Martin S “Tink.” Functional Movement Development 

Across the Life Span. Third Edition. Elsevier. 2012; p. 68–87. 

DOI:10.1016/B978-1-4160-4978-4.00004-1 

37.Martini E, Cesini I, D’Abbraccio J, Arnetoli G, Doronzio S, 

Giffone A, et al. Increased symmetry of lower-limb amputees 

walking with concurrent bilateral vibrotactile feedback. IEEE Trans 

Neural Syst Rehabil Eng. 2021;29:74–84. DOI:10.1109/ 

TNSRE.2020.3034521 

38.Gopalai AA, Lan BL, Gouwanda D. Stochastic resonance for 

enhancing sensory perception: An emerging trend for ADL 

rehabilitation. TENCON 2015 - 2015 IEEE Region 10 Conference, 

2015, pp. 1-5. DOI:10.1109/TENCON.2015.7373098 

39.Collins JJ, Priplata AA, Gravelle DC, Niemi J, Harry J, Lipsitz 

LA. Noise-enhanced human sensorimotor function. IEEE Eng Med 

Biol Mag. 2003;22:76–83. DOI:10.1109/MEMB.2003.1195700 

40.Cochrane DJ. The potential neural mechanisms of acute indirect 

vibration. J Sports Sci Med. 2011;10:19–30. 

 

 

https://doi.org/10.33137/cpoj.v5i1.36744

